Process Capability Study (Oracle Quality Help)


	Previous	Next		Contents	Index	Navigation	Glossary	Library

Process Capability Study

You are the general manager for a precision sheet metal job shop specializing in high quality fabricated parts to the computer industry. Customer orders are usually large and are built-to-order by job based on customer specifications. Each part produced has several critical dimensions that should meet customer specifications. In an effort to meet high quality standards, you have implemented a quality improvement program throughout the shop

Process result for all critical dimensions are collected on the shop floor. For example, one critical dimension is the width of a rear flange that is formed on a press brake. The press brake operator measures the flange width immediately after forming the part. Measurements are randomly taken for four parts produced over the life of the job; for example, the first and last parts produced on the job as well as two from the middle might be measured. The deviation between the specified and actual flange width is calculated by subtracting the actual width from the nominal specification width stated by the specification. Although specifications can vary for each customer, the deviation specification for flange width is 2.0 +/- 0.04 units for all customers.

The immediate goal of your quality improvement program is to maintain a process capability index of at least 1.2. By agreement, process capability is to be estimated only after 40 measurements are recorded. The two Process Capability indices used most often in evaluating the relationship between the process and the customer are Cp and Cpk. They are described mathematically as follows:

Cp measures how dispersed the data is without regard to whether the data is evenly distributed around the midpoint between the Upper and Lower Specification Limits (usually referred to as the nominal value).

Cpk weights the dispersion of data about the nominal which usually lies as a midpoint between the Upper and Lower Specification Limits. The denominator in the Cpk index indicates how much room is needed on either side of the mean to contain most of the distribution. The numerator measures the amount of room from the mean to the specification on the side where there is less room. For both Cp and Cpk, higher values are better and indicate higher process capability.

Cp depends only on the value of the standard deviation and is not affected by the relative position of the mean with respect to the nominal specification. On the other hand Cpk considers the value of the standard deviation and the mean. However, before estimating process capability the issue of statistical control should be addressed. Without statistical control, process capability estimations may prove erroneous due to the inherent instability of the process. Therefore, statistical control should be established before calculating process capability indices.

When a process is in statistical control -- is operating with the minimum amount of variation possible (the variation due to common causes) -- it can be expressed mathematically as follows:

Using the above expression it is possible to calculate process capability using the Rbar value from a Xbar and R chart. Establishing statistical control for flange width became a priority action item. A quality team was set up in the sheet metal job shop. They established three objectives which are:

1. control the process

2. measure process capability

3. improve the process

Setup

Collection Elements

Define the user-defined collection elements Flange Width, Deviation and Machine number in collection element setup.

Collection Plan

Create a collection plan called "Rear Flange" with the following collection elements:

Item

Job Number

Customer Name

Specification

Employee Name

Machine Number

Flange Width

Deviation

Figure 1 - 12 illustrates the process flow for the fabricated parts.

Associated Collection Transactions and Collection Plans

Choose the Transactions button in Collection Plan window. Select the move transaction and define the following collection triggers and conditions:

From Operation Sequence Number <= (at most) 40

To Operation Sequence Number > (greater than) 40

Item = (equals) rear flange

Make results entry for the move transaction mandatory using the Mandatory check box.

Associated Specification Types and Collection Plans

Associate the collection plan with a customer specification using the Specification button in the Collection Plans window. This ensure that you are prompted to select a customer specification when the collection plan is invoked as you enter move transactions.

Profile Option for Specification

The team wants to populate the collection element Specification with the specification value for Flange Width for the customer entered during data entry. This can be achieved by setting the QA:Default Specification Target profile option to Yes. See: Profile Options.

When the default specification value is use, the correct specification for a given customer name is automatically defaulted. The next step is to calculate the deviation of the flange reading from the specification. We use user-defined formula to calculate the deviation. The steps are as follows:

1. From the Collection Plans window, select the 'Flange Width' collection plan element then choose the Actions button. The Quality Actions window appears

2. In the Action Rules region, select the 'is entered' condition.

Assign a value

4. In the Action Details region, choose the Action Details button. The Assign a Value window appears.

5. Select the Deviation collection plan element to Assign the value To.

6. Select the option to assign a value using a formula (this is the default selection).

7. Enter the following formula:

&FW - &SP

where FW and SP are tokens for collection elements Flange Width and Specification respectively

8. Choose the Variables button. The Output Variables window appears.

9. Enter the token name 'FW' for the Flange Width collection plan element. Also enter the token name 'SP' for the Total Quantity collection plan element.

10. Choose OK to exit both the Output Variables and Quality Actions windows.

11. Save your work.

Specification

Every customer has a different set of specifications for Flange width. Therefore for every customer a flange width specification is created in Specification setup. There were seven different customers and therefore a similar number of specifications were created. For the sake of convenience the specifications were provided with their respective customer names.

The quality team is interested in examining the control chart after 40 readings have been recorded in the collection plan.

The collection plan is now ready for collecting data. Every job containing Rear Flange triggers a collection plan where the inspector after selecting the customer specification enters readings on flange width. The team uses the View Quality Results window to see whether at least 40 records have been entered in for the Rear Flange collection plan. After two days, they realize that the requisite number of readings is available for analysis.

The team devises a plan for control chart implementation. They realize that certain considerations are paramount in initiation of a control chart including rational sub-grouping, type of chart, frequency and the type of study being conducted. Their initial conclusion is that the chart should be an Xbar and R chart.

Control Chart Analysis

An examination of the control charts reveals that the process average is not under control. This implies that the average is affected by assignable causes on several occasions:

Three separate points above Upper control limit: point 10,14,17

Three separate points below Lower control limit: point 7, 11, 15

In addition, there is excessive short term variation represented by:

as discussed previously. Even with an R chart in control, the manufacturing process for flange width is not constructed to meet the desired specifications.

As shown below, the pattern suggests that the process average crept upward then was probably adjusted downward. This adjustment cycle, which was repeated for a total of four cycles, is an observation worth investigating.

The inherent process capability was estimated as follows:

and 6 times the standard deviation, which is 0.162, and the specification width is 0.08 which means that the inherent variability is twice what is specified.

Figure 1 - 13

Two major types of investigations are needed to reduce the process variability to the stated specifications:

What are the important assignable causes producing the shifting process average shown by the X bar control chart?

Once identified, how can improvements be effected? The control chart can be continued and watched by production personnel to learn how to control this average.

What are the possible ways of reducing the inherent variability of the process as measured by the standard deviation?

Sometimes relationships between recorded adjustments made in the process and changes in the Range chart can be helpful. For example there is a suggestion that the process variation increased during point 14 to 18 on the R chart. This suggestion would usually be disregarded in routine production; however in a process capability study (which is an industrial investigation of the ability of a process to manufacture within specification) it would warrant investigation.

These two major investigations can lead to other major studies and with every study a greater understanding of process variation takes place. The quality team turns this knowledge into action items and implement them. These action items result in a predictable measure of statistical process control. With every improvement step a SPC chart is generated and process capability is remeasured and the continuous improvement cycle is repeated.